Trajectory for gradient descent algorithms with red and black arrows on... | Download Scientific Diagram
![Gabriel Peyré on X: "Gradient descent is inefficient to find saddle point (Nash equilibrium) for min-max games, because of spiralling behaviour. Beware when training your GANs … https://t.co/KKKGQ4q9JA https://t.co/xjgLuUHPMU https://t.co/hYYH0xWUXP" / X Gabriel Peyré on X: "Gradient descent is inefficient to find saddle point (Nash equilibrium) for min-max games, because of spiralling behaviour. Beware when training your GANs … https://t.co/KKKGQ4q9JA https://t.co/xjgLuUHPMU https://t.co/hYYH0xWUXP" / X](https://pbs.twimg.com/media/DVqXjvFW0AAc3EU.jpg:large)
Gabriel Peyré on X: "Gradient descent is inefficient to find saddle point (Nash equilibrium) for min-max games, because of spiralling behaviour. Beware when training your GANs … https://t.co/KKKGQ4q9JA https://t.co/xjgLuUHPMU https://t.co/hYYH0xWUXP" / X
![optimization - Oscillating around the saddle point in gradient descent? - Artificial Intelligence Stack Exchange optimization - Oscillating around the saddle point in gradient descent? - Artificial Intelligence Stack Exchange](https://i.stack.imgur.com/NyXTy.png)